Skip to content

fix benchmarks for new slice vector syntax #5

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 3 commits into
base: master
Choose a base branch
from
Open

fix benchmarks for new slice vector syntax #5

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
3503d58
fix benchmarks for new slice vector syntax
wolfv Apr 5, 2018
5cc89c1
include 15.05 in range
wolfv Apr 5, 2018
6bd6b96
add manual broadcast loops, remove tabs, fix benchmarks
wolfv Apr 5, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 5 additions & 4 deletions CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ string(TOUPPER "${CMAKE_BUILD_TYPE}" U_CMAKE_BUILD_TYPE)
if (CMAKE_CXX_COMPILER_ID MATCHES "Clang" OR CMAKE_CXX_COMPILER_ID MATCHES "GNU" OR CMAKE_CXX_COMPILER_ID MATCHES "Intel")
CHECK_CXX_COMPILER_FLAG("-std=c++14" HAS_CPP14_FLAG)
if (HAS_CPP14_FLAG)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Ofast -ffast-math -march=native -std=c++14 -pthread")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Ofast -ffast-math -march=native -std=c++14 -pthread -Wno-narrowing")
else()
message(FATAL_ERROR "Unsupported compiler -- xtensor requires C++14 support!")
endif()
Expand Down Expand Up @@ -150,9 +150,10 @@ endif()
message("Found xsimd : ${xsimd_INCLUDE_DIRS}\n\n")

if (EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/src/profile_snip.cpp)
add_executable(profile_snip
src/profile_snip.cpp
${XTENSOR_HEADERS})
add_executable(profile_snip
EXCLUDE_FROM_ALL
src/profile_snip.cpp
${XTENSOR_HEADERS})
endif()

add_custom_target(xbenchmark
Expand Down
118 changes: 94 additions & 24 deletions src/benchmark_broadcasting.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@
#include "xtensor/xrandom.hpp"
#include "xtensor/xtensor.hpp"
#include "xtensor/xarray.hpp"
#include "xtensor/xfixed.hpp"

#ifdef HAS_PYTHONIC
#include <pythonic/core.hpp>
Expand All @@ -27,37 +28,106 @@

namespace xt
{
void xtensor_broadcasting(benchmark::State& state)
{
using namespace xt;
using allocator = xsimd::aligned_allocator<double, 32>;
using tensor3 = xtensor_container<xt::uvector<double, allocator>, 3, layout_type::row_major>;
using tensor2 = xtensor_container<xt::uvector<double, allocator>, 2, layout_type::row_major>;
void xtensor_broadcasting(benchmark::State& state)
{
using namespace xt;
using allocator = xsimd::aligned_allocator<double, 32>;
using tensor3 = xtensor_container<xt::uvector<double, allocator>, 3, layout_type::row_major>;
using tensor2 = xtensor_container<xt::uvector<double, allocator>, 2, layout_type::row_major>;

tensor3 a = random::rand<double>({state.range(0), state.range(0), state.range(0)});
tensor2 b = random::rand<double>({state.range(0), state.range(0)});
tensor3 a = random::rand<double>({state.range(0), state.range(0), state.range(0)});
tensor2 b = random::rand<double>({state.range(0), state.range(0)});

for (auto _ : state)
{
tensor3 res(a + b);
benchmark::DoNotOptimize(res.raw_data());
}
}
BENCHMARK(xtensor_broadcasting)->RangeMultiplier(MULTIPLIER)->Range(RANGE);
{
tensor3 res(a + b);
benchmark::DoNotOptimize(res.raw_data());
}
}
BENCHMARK(xtensor_broadcasting)->RangeMultiplier(MULTIPLIER)->Range(RANGE);

void xarray_broadcasting(benchmark::State& state)
{
using namespace xt;
using allocator = xsimd::aligned_allocator<double, 32>;
using tensor3 = xarray_container<xt::uvector<double, allocator>, layout_type::row_major>;
using tensor2 = xarray_container<xt::uvector<double, allocator>, layout_type::row_major>;

tensor3 a = random::rand<double>({state.range(0), state.range(0), state.range(0)});
tensor2 b = random::rand<double>({state.range(0), state.range(0)});

for (auto _ : state)
{
tensor3 res(a + b);
benchmark::DoNotOptimize(res.raw_data());
}
}
BENCHMARK(xarray_broadcasting)->RangeMultiplier(MULTIPLIER)->Range(RANGE);

template <std::size_t N>
void manual_broadcast_xtensorf(benchmark::State& state)
{
auto a = xt::xtensorf<double, xt::xshape<N, N, N>>();
auto b = xt::xtensorf<double, xt::xshape<N, N>>();
for (auto _ : state)
{
auto c = xt::xtensorf<double, xt::xshape<N, N, N>>();
for (std::size_t i = 0; i < a.shape()[0]; ++i)
for (std::size_t j = 0; j < a.shape()[1]; ++j)
for (std::size_t k = 0; k < a.shape()[2]; ++k)
c(i, j, k) = a(i, j, k) + b(i, j, k);
benchmark::DoNotOptimize(c.raw_data());
}
}
BENCHMARK_TEMPLATE(manual_broadcast_xtensorf, 3);
BENCHMARK_TEMPLATE(manual_broadcast_xtensorf, 8);
BENCHMARK_TEMPLATE(manual_broadcast_xtensorf, 64);

void manual_broadcast_xtensor(benchmark::State& state)
{
auto a = xt::xtensor<double, 3>::from_shape({state.range(0), state.range(0), state.range(0)});
auto b = xt::xtensor<double, 2>::from_shape({state.range(0), state.range(0)});
for (auto _ : state)
{
xt::xtensor<double, 3> c = xt::xtensor<double, 3>::from_shape({state.range(0), state.range(0), state.range(0)});
for (std::size_t i = 0; i < a.shape()[0]; ++i)
for (std::size_t j = 0; j < a.shape()[1]; ++j)
for (std::size_t k = 0; k < a.shape()[2]; ++k)
c(i, j, k) = a(i, j, k) + b(i, j, k);
benchmark::DoNotOptimize(c.raw_data());
}
}
BENCHMARK(manual_broadcast_xtensor)->RangeMultiplier(MULTIPLIER)->Range(RANGE);

void manual_broadcast_xarray(benchmark::State& state)
{
auto a = xt::xarray<double>::from_shape({state.range(0), state.range(0), state.range(0)});
auto b = xt::xarray<double>::from_shape({state.range(0), state.range(0)});
for (auto _ : state)
{
xt::xarray<double> c = xt::xarray<double>::from_shape({state.range(0), state.range(0), state.range(0)});
for (std::size_t i = 0; i < a.shape()[0]; ++i)
for (std::size_t j = 0; j < a.shape()[1]; ++j)
for (std::size_t k = 0; k < a.shape()[2]; ++k)
c(i, j, k) = a(i, j, k) + b(i, j, k);
benchmark::DoNotOptimize(c.raw_data());
}
}
BENCHMARK(manual_broadcast_xarray)->RangeMultiplier(MULTIPLIER)->Range(RANGE);

#ifdef HAS_PYTHONIC
void pythonic_broadcasting(benchmark::State& state)
{
auto x = pythonic::numpy::random::rand(state.range(0), state.range(0), state.range(0));
auto y = pythonic::numpy::random::rand(state.range(0), state.range(0));
void pythonic_broadcasting(benchmark::State& state)
{
auto x = pythonic::numpy::random::rand(state.range(0), state.range(0), state.range(0));
auto y = pythonic::numpy::random::rand(state.range(0), state.range(0));

for (auto _ : state)
{
pythonic::types::ndarray<double, 3> z = x + y;
benchmark::DoNotOptimize(z.fbegin());
}
}
BENCHMARK(pythonic_broadcasting)->RangeMultiplier(MULTIPLIER)->Range(RANGE);
{
pythonic::types::ndarray<double, 3> z = x + y;
benchmark::DoNotOptimize(z.fbegin());
}
}
BENCHMARK(pythonic_broadcasting)->RangeMultiplier(MULTIPLIER)->Range(RANGE);
#endif

}
Expand Down
Loading